An experimental evaluation of Kalman filtering

Abstract

The effectiveness of the stationary form of the discrete Kalman filter for state estimation in noisy process systems was demonstrated by simulated and experimental tests on a pilot plant evaporator. The filter was incorporated into a multivariable, computer control system and resulted in good control despite process and/or measurement noise levels of loyo. The results were significantly better than those obtained when the Kalman filter was omitted or replaced by conventional exponential filters. In this application the standard Kalman filter was reasonably insensitive to incorrect estimates of initial conditions or noise statistics and to errors in model parameters. The filter estimates were sensitive to unmeasured process disturbances. However this sensitivity could be reduced by treating the noise covariance matrices R and Q as design parameters rather than noise statistics and selecting values which result in increased weighting of the process measurements relative to the calculated model states.

Comparison of the Kalman filter with classical digital filter

“…Thus in the example of Hamilton et al (1973) even the best esponential filter, which could not compensate for system dynamics, led to oscillations in the tight optimal feedback control loop. As pointed out however in the original work (Hamilton, 1972), loosening of the loop with a lag designed controller might improve control significantly although of course not to the optimal level.…”

“…The availability of this additional information can be an important factor in filter selection. For example, use of the exponential filter in the multivariable control system of Hamilton et al (1973) necessitated generation of one of the state variables by separate means.…”

State estimation and state feedback control for continuous fluidized bed dryers

“…For the observer design, diagonal Q and R matrices were assumed with each matrix having equal diagonal elements such that (Hamilton et al, 1973);…”

“…The most popular estimation technique for a continuous process is Kalman filter. Since the publication of the classic papers by Kalman (1960) and Kalman & Bucy (1961), the application of the Kalman filter and its extended design to chemical processes has received extensive study (e.g., Dimtratos, Georgakis, El-Aasser, & Klein, 1988;Hamilton, Seborge, & Fisher, 1973;Seinfeld, 1970;Quintero-Marmol, Luyben, & Georgakis, 1991;Tatiraju & Soroush, 1997). In drying control, reliable sensors have been constructed for direct measurement of material moisture content, but they are restricted to a small class of agricultural products such as grains (e.g.…”

Fault detection and diagnosis in a closed-loop nonlinear distillation process: application of extended Kalman filters

“…Though there are a few papers dealing with closedloop systems, it is not clear whether fault detection and diagnosis can be successfully performed based on closedloop performance. Hamilton et al (1973) applied the Kalman filter for state estimation in a closed-loop pilot plant evaporator. The filter estimates were sensitive to unmeasured process disturbances (faults), since they were not accounted for by the linear process model.…”